Someone suggested an interesting scenario where young people would show up to vote way more than usual. I decided to go one step further and estimate the electoral results if only young (under 25) people voted or, the exact opposite, if only people over 65 showed up. These are extreme scenarios, but I think it's always interesting to try to see the extreme first. The reality is then a mix of these extreme cases.

1. Only young people.

Using the latest Ekos poll, I did the following manipulation: for each party, I compared the level of support among everybody to the one among young voters only. For instance, for the NDP, these numbers are respectively 24.7% and 28.8%. Therefore, young voters tend to vote more for the NDP in a proportion of 1.17, or 17% (note: to be really correct I should have compared the level of support of the over 25 to the under 25. The problem is that we only have the number of observations in each age-ranges, not the sampling weights. So I can't do the exact calculations as Ekos. In any ways, the number of under 25 isn't that big, therefore not affecting the calculations much).

Therefore, I increased the percentages for the NDP in every province by 17%. Of course, this assumes that there isn't province-young specific effect. It's an approximation, but it's the best I can given the sample size and the information provided.

For the Conservatives, the support under 25's is only 79% of the overall support. For the Liberals, the ratio is 58.7% (!). Finally, the Green enjoyed a level of support 2.46 time the national average! And for the Bloc, it's 30% higher.

The results? Here they are:

You may be surprised to see the Conservatives still winning, but remember, even by being down 21%, the Conservatives would still be first in many provinces. As for the NDP, the problem for it is that the Bloc also enjoys more support among young people, thus allowing it to sweep Quebec. The party that would be really shaken is the Liberals. I didn't know their level of support was so low among the under 25. The only reason they are not even lower is because they still benefit from some huge margins during the last election. It's a simple simulation exercise, I'm well aware of the various limitations.

2. Only senior (65+).

Same mechanic, but this time with the seniors only. This time around, the Conservatives are up 24%, the Liberals up 23%, the NDP down 25%, the Green down 62% and the Bloc down 38%.


Quite a contrast with the previous scenario. Please notice that you should compare these scenarios to my "latest projections" on the right, not the Ekos poll only. So for the CPC, they would gain 27 seats with this scenario 172 instead of 145).

Overall, what I like about this exercise is to show how different the results could be. What this means in real life is that because young people tend to vote so little (while the turnout among seniors is quite high), the overall results is closer to what older people want. So you can show this post to your friends who think that voting doesn't make any difference and try to convince them to vote. As explained also here, if young people voted more, that could potentially make a big difference.

It's a beautiful Saturday here in Vancouver and I won't be working this afternoon. However, I have a few ideas for some posts, such as: what if the Liberals Party didn't exist? Or another one where I try to find where the orange wave comes from (specifically, I simulate how much difference it makes whether the orange waves comes from some Liberals voters switching to the NDP or from some Bloc voters).

Do you have any idea or request like this? I had one earlier today asking me to simulate a higher turnout of young people. I'll do that later today.

So if you would like me to do any simulation like this, lemme know. I'll tell you if I can indeed do that.

I thought of something tonight: since the NDP is rising in Quebec and clearly getting votes from the Bloc, then we should see something happening among the remaining Bloc voters regarding their second choice, no? I mean, if you look at this Ekos poll from early April (when the NDP was at only 18% and the Bloc at 33%), you see that 42% of Bloc voters have the NDP as second choice. Fast forward to this poll, released yesterday. You still have 40.1% of Bloquistes stating the NDP as their second choice. How do you explain that? Isn't it likely that the people who switched were more likely to have the NDP as second choice already? Therefore, the remaining Bloc voters should have different second choices than before.

I just thought I would ask some of my regular commenters to see if I'm the only who finds that a little bit odd.

With the recent NDP's surge, I need to revise my projections based on the average of most recent polls. After all, the NDP pointing around 30% in Quebec has been shown not in 1 poll, not in 2, but in 4 (Ekos, Ipsos, Forum and Crop). And I'm sure Nanos will soon show the same things, as soon as their 3-days rolling panel is replaced with new interviews. Therefore here you go, these are the new latest projections.

Let's start with the "star" of the moment: the NDP. They are up to a all-time high of 47 seats, mostly thanks to the 8 seats in Quebec. And if the trend observed in the last 4 polls keep going, they will soon be projected to win way more than that. By the way, 4 polls provide us with quite accurate results. Indeed, using only these four polls, we would get the following margins of errors in Quebec (note: I'm "giving" MOE to the Crop poll even though it was conducted online and is thus not a perfectly random sample. I honestly don't think it's such a big deal to be conducted online as Angus-Reid has been doing that for years and they have been very accurate):

MOE for the average of four polls: ±2.29%

CPC: 18.5%

LPC: 16.3%

NDP: 32.7%

Green: 3.2%

Bloc: 27.6%

So as you can, even by applying the MOE, we still get the NDP above 30%! At 95% confidence intervals, you can thus exclude the NDP to be below this threshold. Using this average only, we would get 26 NDP seats in Quebec and 26 Bloc. For the country, the NDP would finish close third with 66 seats, while the Conservatives are mostly unaffected (at 145 seats) since the NDP's gains are from Quebec. I'm not confortable enough to use these four polls only for the "official" projections yet. I provide this information out of curiosity, that's all.

So we have two main questions regarding the NDP: will the surge last in Quebec? And secondly, will it spread elsewhere? In 2006, most of the gains for Harper came from Quebec, so it isn't impossible that the rise will be contained in one province. I lived in Quebec and I have observed its politics for years. It's a very particular place where waves and movements happen really fast. In 2006, Harper started the campaign at 5% and finished at 25% with 10 MPs. In 2007, the ADQ rose from 4 MLAs and 15% to 31%, 39 MLAs and almost won the government! So can the NDP really get a lot of new MPs there? Totally! Of course, they will now be under attacked from the Bloc on a daily basis but you only have one week to go. Besides, it's easy for Gilles Duceppe to scare Quebecois off with things such as "abortion", Bush-like" or "oil sands" when he has to fight the Conservatives. But it's a all other story when he has to fight the NDP who shares so many policies with the Bloc. In many ways, the Bloc is a separatist NDP (I know I'll upset some die hard Bloc fans here, but this is the truth). The only card the Bloc can play is that the NDP is likely too "centralisateur" but I'm not sure this is the high issue in Quebec right now.

The other parties except the Bloc are not affected (yet) by the NDP's surge. Even though this party is clearly up in recent polls outside of Quebec, it isn't big enough to cause majors changes. I'm pretty sure Harper is not too worried at the moment as it doesn't really change his game plan for a majority. He is where he was 1 week ago: he needs 10 seats. On the other hand, Michael Ignatieff must be wondering what to do. If he spends the last week attacking the NDP, he could potentially leave Harper alone and lose some ridings in the GTA. This strategy would make sense though as this is what they have been trying to do since the beginning: getting vote from the NDP. It might be too late though.

You can see the riding-by-ridings pdf here. If you look in Quebec, you can see the NDP is winning a lot of close races, but is also losing quite a lot! What this means is that the NDP is clearly in the "paying zone" in this province. Every additional percentage point (especially if taken from the Bloc) could result in many new seats. I'd pay big money to see the partisans of the Bloc trying to change their startegy at the last moment (I'm sure they never planned on having ads against the NDP!).

The next couple of days will tell us if we can really expect this orange wave on May 2nd or not.

If you have twitter and are following people interested in politics, you know by now that the NDP's surge is real and really crazy. In particular, the recent polls show the NDP first in Quebec. Deniers will hide behind the fact that the Crop poll was conducted online. That kinda makes me laugh as we never heard complaints about polls conducted with an online panel before. In particular, where were all these statisticians when Angus-Reid or Crop were showing the NDP at 15% and the Bloc at 38%? Anyway, Crop is not the only one showing the same thing, we see a big increase in Quebec for the NDP in Nanos, Forum Research (I'm waiting for the details of their poll, as they gladly accepted to provide it to me last time) and Ekos.

Also, if you think the surge will go away as the NDP voters are not committed, I think you are wrong. Most polls show that people voting NDP are not more likely to change their vote than the Tories voters. What might happen though is that now that Jack Layton and the NDP are rising, they will be attacked from all sides. We'll see how well they can resist. Having a candidate in Punta Canada instead of running in his riding is one thing where you are low, but these kind of "scandals" have more impacts when you are the focus of attention.

In any ways, using only the latest polls from Nanos, Crop (in Quebec), Ekos and Forum (last two show the NDP tied nationally with the Liberals!), we get the following projections. Yes that's right, the NDP would have a real shot at being the official opposition. And maybe the government? (i.e: a scenario ala Ignatieff...). If I use only EKos, Forum and Crop, the NDP would even go as high as 76 seats, 4 more than the Liberals. The key in Quebec for the Bloc is really around 30%. If they fall below that (and the NDP gets above it), then the Bloc is starting to lose seats really fast. Here are the riding-by-riding projections.

Also note that the NDP is rising everywhere, especially in the Atlantic and the Prairies. They are also up in BC. Actually, the only region where they haven't experienced some kind of boost is Ontario. And this is what is currently saving the Liberals. If Ontarians were to switch to the NDP, Michael Ignatieff could well finish below 60 seats...

Of course, it's only 4 polls. But hey, 4 is better than 1. You can't deny there is a trend going on, especially in Quebec. We have about one week and half left to this campaign and a lot of things can happen.

A note about the Conservatives. In Quebec, they aren't that affected by the NDP. Actually, if the NDP was to score around 20%, they could even be helped by a splitting of the votes. But even if the NDP sweeps Quebec and gets more than 30%, the Tories would likely keep 5-7 seats. In the rest of Canada, as opposed to what a lot of people think, the NDP'surge isn't all good news. Sure it could give them 2 ridings in Ontario, but in the rest of the country, the NDP would actually steal seats from them. This is especially the case in the Prairies and BC.

Note to my English readers: since I firmly believe in a bilingual Canada and since this post is about Quebec, I decided to write it in french. You can use Google translate if you really have to, or better, try to read it anyway. Posts in french will remain the exception.

Le dernier sondage Crop nous indique quelque chose de totalement fou: le NPD serait en tête au Québec. La vague orange semble ne plus vouloir s'arrêter. Les Québécois semblent souvent être de même: ils ne considèrent même pas voter pour un parti (comme le PCC en 2004 ou l'ADQ avant 2007) et quand ce parti commence à grimper, alors là plein de gens se joignent au mouvement.

Selon le sondage, le NPD serait à 36%, le Bloc à 31%, le PLC en chute libre à 13% et le PCC plus ou moins stable à 17%. En termes de sièges, cela donnerait:

PCC: 7

PLC: 9

NPD: 31

Bloc: 28

Incroyable! Le Bloc serait gravement touché. Voici le pdf comté-par-comté.

Bien sûr, il ne s'agît que d'un sondage, mais la tendance est là et elle est très lourde. Non seulement le NPD serait-il en tête au Québec, mais il aurait une réelle chance de former l'opposition officielle!

[MàJ] Voici un sondage Ekos montrant le NPD premier également. La différence étant que le PLC est bien plus élevé dans ce sondage et le Bloc croupie à seulement 23.7%! En faisant la moyenne des deux sondages, cela donnerait 32 NPD, 22 Bloc, 12 Libéraux et 9 Conservateurs...

The surge of the NDP in the polls is clearly translated into additional seats, mostly in Quebec. If the Bloc keeps falling (and depending on where the swing comes from for the NDP in this province) Jack Layton could potentially get as many as 7 MPs. It's really bad news for the Bloc. Until now, they were only facing competition from center-to-right-wing federalist parties. Therefore they could capitalize on the left-wing AND the separatist vote. Now that the NDP is a real player in Quebec, the Bloc is basically left with the separatist vote only and during a federal election, it's harder to get this vote out. Please note that the model in Quebec for the NDP is now a mix of estimation/calibration. Basically, as long as the swing for the NDP is not completely uniform, they can expect 4-7 seats. You really have many ways to achieve that: increase in the North and South suburbs of Montreal; increase in Montreal-West; increase at the expense of the Bloc, etc. Again, anything but a completely flat and uniform swing will give them more than 2 seats.

For the Conservatives, if it wasn't for a couple of bad polls in the Atlantic, they would be in majority territory. However, they have 128 safe seats and are involved in 40 close races. Therefore, the maximum is large for them with 168 seats. By the way, these projections are made with the improved model for Ontario where another factor has been added: the swing experienced by parties in close wins and close losses. This explains why the Tories are now at 54 seats and much closer to the Grits in a lot of the GTA ridings.

For the Liberals, honestly, you need to work really hard to find some good news. I think it's clear that the strategy of Michael Ignatieff didn't work so far. The idea of solidifying the party to its left, by getting votes from the NDP, and therefore forcing Harper to another minority, kind of backfired. Not only the Liberals are not stealing votes from the Tories, but they are now threaten by the NDP. We still have two weeks left, but the momentum is clearly behind the NDP. And with Ignatieff confirming he would be ready to bring Harper down and form the government with (informal) support of the Bloc and NDP, I don't see this trend changing. If anything, it will convince some undecided voters to vote for stability and election-free period of 4 years.

For the Bloc, thanks to the electoral system, they would probably still win 45 seats, even though they represent only 32% of the votes... However, now that the orange wave is really started in Quebec, we'll see how the last two weeks will turn out. I think the big threshold for them is to stay above 30%. If they fall below that, it could get really ugly for them.

The riding-by-riding projections are here.